THE INTERACTION OF BINARY TROPICAL CYCLONES IN A COUPLED TROPICAL CYCLONE-OCEAN MODEL

Alexander Khain (Hebrew University of Jerusalem, Israel)

Isaac Ginis (GSO/URI)

and

Alexander Falkovich (NCEP/NOAA)

Introduction

Two or more tropical cyclones existing simultaneously interact with each other when the separation distance becomes less than about 1450 km (Brand 1970). These situations more frequently occur in the western and eastern North Pacific (Ramage, 1972; Lander and Holland, 1993). Several hurricanes sometimes develop simultaneously over the Atlantic Ocean too as it was observed, for example, in August 1995.  The interaction of tropical cyclones frequently causes sharp changes of their tracks and translation speed.  Large forecast errors can be associated with an incorrect assessment of these situations (Brand, 1970; Neumann, 1981). The binary vortices can merge or move away depending on the storm structures, intensities and the separation distance.  Dong and Neumann (1983) found that the distances between storms, initially separated by less than 900 km, decreases with time in 60% of cases. During the mutual approach, one member of the interacting pair usually decays and loses its identity at relatively large distances (on the order of several hundred kilometers) from the survived (winner) vortex. In other cases, however, attraction of binary storms may sharply changes to repulsion after a certain separation distance is being reached (Lander and Holland, 1993). Thus, different scenarios are possible which determine the final result of interaction between binary tropical storms.

In this project we investigate the motion and evolution of binary tropical cyclones using a coupled tropical cyclone-ocean movable nested grid model. The model comprises of eight-layer atmospheric and seven-layer ocean primitive equation models. Condensation heating is calculated at resolvable grid scales so that cumulus convection is hydrostatic but explicit.  In a set of numerical experiments, pairs of axisymmetric weak vortices of both equal and unequal intensity and size were initially separated by specified distances.

• Complete merger (CM) is the coalescence of the storms into a single storm without any loss of their vorticity into surrounding environment
• Partial merger (PM) defines the interaction when one of the storms is only partially entrained and captured by the opposite storm.
• Straining out (SO) is when the weaker storm is destroyed by the stronger one at some separation distance
• Mutual straining out (MSO) is when both storms are stretched out and weaken at some separation distances. At some separation distance the storm approach can be replaced by storm repulsion.
• Elastic interaction (EI) is when the storms interact without any significant changes in their intensity and structure.

At separation distances of 640 km the interacting storms experienced partial merger (PM). At intermediate (700 km to about 1000 km) initial separation distances two regimes of storm interaction have been found: straining out (SO) characterized by complete disintegration of the weaker storm and mutual straining out (MSO) characterized by weakening and dissipation of both storms. SO occurred when the interacting storms had substantially different intensities and strengths. MSO was observed when the interacting storms were comparable in size and intensity. In the latter case, the storms were unable to approach each other at distances smaller than a certain minimum distance (of about 450-500 km) without being mutually stretched out. Moreover, initial attraction of the storms in this regime was replaced by repulsion, in agreement with observations (Lander and Holland, 1993). One of the possible causes hindering further storm attraction is the displacement of the maximum latent heat release to the opposite sides of the interacting storms. The storms can be pushed away from each other due to the tendency of tropical cyclones to displace toward the areas of maximum heating.

The type of interaction depends on comparable strength of the storms in a pair. The storm strength, in its turn, depends on various factors such as the Coriolis force, SSTs, vertical and horizontal shears of the background flow. We found that the result of storm interaction  also  depends on initial location of the storms. In our experiments, the storms develop from vortices initially the latitude of 15N. The storm initially located to the west (storm W) has an advantage over the storm initially located on the east (storm E): the latter storm moves faster northward and turns out to be weaker under other conditions being equal. Thus, the result of storm interaction is depended on what storm (eastern or western) was stronger initially. Note that comparably small changes in structure and strength of interacting storms can lead to different scenarios of their interaction. This result implies that forecasting the result of binary storm interaction  is  rather difficult.

The results of a series of sensitivity experiments with different convective parameterization illustrated the importance of adequate simulation of the storm structure for predicting the results of storm interaction.  In the conducted experiments with a CISK-parameterization of convective heating in the way similar to that used by Wang and Holland (1995) the storms were nearly axisymmetric, very compact and continued approaching each other until they merged. Thus, the type of storm interaction  depends dramatically on  the way convective heating is described.  It clearly indicates the importance of utilization  of  realistic convective parameterization.

The ocean coupling may significantly affect the binary storm interaction. The storm-induced SST decrease results in a reduction of storm intensity, slower mutual orbiting and, therefore, substantially different tracks of binary storms. The changes in storm structures due to ocean coupling also cause the decrease of the MAS. The ocean coupling may also change the interaction regime. One of the storms, moving over the cold wake created by the other, can significantly weaken and get destroyed by the stronger counterpart. Thus the ocean coupling may be crucially important in determining which of the storms will be the winner during the storm merger or straining out.

The regimes of binary storm interaction must also depend on the structure of the background flow. Analyses of the environmental effects will be the subject of our future investigation.